USB support for RMI processors

[M. Warner Losh]

In message: <eaa228be1003012057nd382421vf4386953542243dc at mail.gmail.com>
            Juli Mallett <jmallett at freebsd.org> writes:
: On Mon, Mar 1, 2010 at 20:53, C. Jayachandran <c.jayachandran at gmail.com> wrote:
: > The userland compiles with the patches and with -DNO_USB
: > -DNO_BLUETOOTH (using ld to convert binary to n32 obj fails - needs to
: > look at this). But the main battle will be ahead, the syscall,
: > exception and pobably signal handling and executable support needs to
: > be fixed before init goes thru.  I'm working on this.
: 
: I've made a hackish change to syscall stuff that works well enough but
: breaks o32 support; it should be obvious how to fix that:
: 
: http://svn.freebsd.org/viewvc/base/user/jmallett/octeon/sys/mips/mips/trap.c?r1=204399&r2=204534&sortby=date

I think this is why we'll need to know the ABI that the binary is
running :)

: I'm fighting with rtld right now and believe I know the source of my
: misery but my tree has stopped working for some unrelated reason so
: now I'm trying to figure out what I messed up.

__start and rtld is very intimately linked.  And both are sensitive to
the ABI.  I have some saved patches in my tree that I've not had a
chance to test...

: BTW I've found it very useful to work with WITHOUT_DYNAMICROOT while I
: worked on the issues more fundamental than the rtld problem.  Trivial
: signals seemed to work fine.

Yea, me too.  :)

Warner